CREATING 3D MODELS FOR AIRCRAFT COMPONENTS AND SYSTEMS FOR SUPPORT... AIRCRAFT ACCIDENT INVESTIGATION

advertisement
Robertson, Gary
CREATING 3D MODELS FOR AIRCRAFT COMPONENTS AND SYSTEMS FOR SUPPORT IN
AIRCRAFT ACCIDENT INVESTIGATION
Gary Robertson
ShapeQuest Inc.
Gary@shapequest.com
Working Group V/1
KEY WORDS: Laser targeting, binary target generation, automatic target reading, digital imagery, 3D models,
reflectorless total stations, digital image matching, texture mapping, feature extraction, accident investigation, aircraft
components, wiring, calibration, bundle solution.
ABSTRACT
This paper describes the procedures used in 3D modeling of aircraft wire bundles and components, including cockpit
and interior details of a wide body aircraft. A combination of several methods are used in the data collection and
modeling. Systems used are servo controlled reflectorless total stations and digital photogrammetry. The paper
describes the photogrammetric targeting techniques using projected binary targets and projected laser patterns. The
automated feature extraction, and automated target reading procedure used to create the 3d models will be described.
The paper will also discuss time comparisons between the various methods and time to 3D models. Also discussed are
the problems of working in the interior of aircraft and the stabilization problems inside the aircraft. Several examples
are discussed and illustrated from small controls, instruments and major interior components of the aircraft that were
measured and modeled in and outside of the aircraft.
1 INTRODUCTION
Photogrammetry has been used for the last two decades in aerospace manufacturing and quality assurance. The
procedures remain somewhat the same, with a drastic change in equipment and 3dmodelling capabilities. Although
automated target reading and has been around for some time as reported by (Robertson, Wyatt, 1984), (Robertson,
1986) it required large format photogrammetric cameras and expensive precession scanners and computing power. Now
with smaller hand held digital cameras and new total station technology and expanded software capabilities now offers
not only first order accuracy but also 3d modeling in near real time.
1.1 Aircraft Interiors
The problem of working in a large aircraft during assembly or inspection is movement with numerous work groups
walking in the aircraft. This obviously presents a stability problem not so much for the acquiring the images but for the
survey and any type of target projection. This particular project required us to create 3d models of wiring bundles in a
wide body jet, in addition to develop 3d models of the cockpit and interior elements. Some of these interior elements
were recorded outside the aircraft. For the photogrammetric recording of the interior several approaches were used.
Convergent digital images would be used for the wire bundle and use of projected target patterns for automated target
measurement. This would also augment the data acquired with the survey procedure with the total station. For the
projected target pattern we used a laser and a projected precision dot pattern that we have used with a modified raster
technique as reported in (Robertson, 1988)
1.1.1 Survey Control. The instrument used in the aircraft was a Leica TCRM1103, a motorized 3-second total station,
featuring Leica's unique visible red laser EDM. One of the very significant advantages of the visible red laser used by
Leica is the small spot size (approximately 8X24mm) that makes it possible to use the spot itself for aiming the
theodolite at target points (Figure 1 Figure 2). This proved to be a very important feature, especially in the cockpit area,
were a lot of points were well under 1m from the instrument. Since this distance is well under the minimum focussing
distance for nearly all-total stations, the visible red laser allowed us to make measurements that would not have been
possible otherwise, due to an inability to focus on the target point. The small spot size also minimized the impact of
adjacent surfaces on the distance measurements to the target points.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
677
Robertson, Gary
Figure 1
Figure 2
We used Leica's Local Resection, Free Station and Orientation & Height Transfer applications, running onboard the
TCRM1103, to facilitate setting the instrument into the aircraft coordinate frame of reference initially, and as we moved
the instrument around the aircraft. The applications automatically used the motors of the instrument to drive to the third
and successive points during the orientation processes, and even inverted the telescope and re-pointed to all the target
points, allowing easy measurements in both instrument faces. The short distances involved did not warrant the accuracy
potential of inverting the telescope, but doing so allowed us to better cope with the continual motion of the aircraft as
we worked.
Figure 3
Scanning of one of the wall panels in the first class cabin was performed using Leica's Face Scan application. Face Scan
allowed us to measure thousands of points to develop an accurate digital model of the panel, and it performed the
measurements unattended. All that was required was to switch the instrument off at the end of the measurements. These
same panels were also measured using an automated photogrammetric procedure.
1.2
Photogrammetric Measurement and Modeling the Interior of the Aircraft The measurement and modeling
phase of the aircraft interior was accomplished using ShapeCapture software. The ShapeCapture software provides for
accurate 3D measurement, full camera calibration and modeling and automatic target extraction, matching, and
measurement. For a portion of the photogrammetric work automated target extraction, matching, and measurement was
used. We found that the laser-projected points were in most case’s poor. The problem was the illumination and the
number of points that could be used with the power requirements of the laser. Figure 4 illustrates the projected laser on
the interior wiring. The raster steel slide was also used it produces binary targets, we use various combinations where
one slide can project up to 5,460 points on the object see Figure 5.0. An additional slide format projects 1,395 points
with a larger diameter. 3D models were created of the overhead wire bundles, (Figure 6) cockpit and other key interior
features of the aircraft. Figures 7 to 11.
678
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
Robertson, Gary
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
679
Robertson, Gary
Figure 10
Figure 11
In addition interior modules were imaged outside the aircraft such as galleys and lavatory module. The ShapeCapture
program provided 3d models that were placed in MicroStation Cad program for indexing. In addition full texture
mapped models were created in ShapeCapture as shown in Figures 12, 13. The Figures 14, 15, 16, 17 illustrate the
models that were indexed and rendered inside of MicroStation.
Figure 12
Figure 14
680
Figure 13
Figure 15
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
Robertson, Gary
Figure 16
2.0
Figure 17
Automated Measurement and Modeling
The interior panel that was measured with the Leica TCRM1103 (Figure 3) was also measured outside the aircraft
(Figure 19) along with other parts with The ShapeCapture program.
Figure 18
Figure 19
Figure 18 illustrate the procedure of the projected dot pattern on the surface with the camera positions shown. Figure 19
shows the window panel with the projected pattern. ShapeCapture reads the projected targets automatically and using
stereo matching computes the 3d coordinate geometry and then allows you to assign it a set for automated modeling of
the data. Figure 20 and 21 illustrates the 3956 points and 4339 target points being measured and matching 3666 points
in seconds.
Figure 20
Figure 21
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
681
Robertson, Gary
The completed model is illustrated in Figure 22. A modified version could yield accurate 3d measurement and matching
in near real time.
Figure 22
3
Conclusions
With the advancement of new automated systems, utilizing the latest in motorized laser total stations, and digital
cameras and photogrammetric software with automated measurement and matching, reduces the field time and data
acquisition time substantially. The most important aspect of this is the modeling, and CAD side with the abilities of
accurate surface fitting and texture mapping. During the earlier nineteen eighties we developed a system to measure
targeted points automatically and with great accuracy but could not created 3d models. It was expensive and slow as to
the standards of today. We now have the capabilities of first order type accuracy with full accurate 3D modeling
automatically.
ACKNOWLEDGEMENTS
I would like to thank Charles Lee of Leica for his assistance and support
REFERENCES
Robertson, G., A., Wyatt, 1984 “Real Time Image Processing and Target Recognition System for Photogrammetric
Mensuration”, presented paper XV ISPRS Congress Rio de Janeiro Brazil
Robertson, G. 1986 “Test of Photogrammetric Accuracy for Mensuration of Aerospace Tooling” presented paper
ISPRS Commission V symposium Ottawa, Canada
Robertson, G. 1988 “Mensuration of Body Shapes Using an Automated Photogrammetric Approach” presented paper
Commission V ISPRS 16th Congress Kyoto, Japan
Robertson, G., 1990, “Aircraft Crash Analysis Utilizing a Photogrammetric Approach” ISPRS Commission V
symposium Zurich, Switzerland
Robertson, G., 1994 “Accident and Crime Scene Analysis Utilizing Real Scene Photogrammetry” ISPRS Commission
V symposium Melbourne Australia
682
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
Download